Lying toward AI agent: the roles of type of lie, moral disengagement and perceived emotional ability

Purpose This paper aims to investigate the effect of artificial intelligence (AI) agent on consumers’ lying behavior. It provides novel insights into the pivotal role of lie type (material egoistic lie versus social egoistic lie) in shaping consumers’ lying behavior toward the AI agent (versus human), and the mediating roles of moral disengagement and perceived emotional ability for each type of lie. Design/methodology/approach Seven studies test the proposed theoretical framework. Based on the results of a pilot study that established the classification of lie (material egoistic lie versus social egoistic lie), six experimental studies using different designs (including an incentive-compatible study) were conducted to test the hypotheses. Study 1 (1a and 1b) and Study 2 (2a and 2b) tested the direct effect of the AI agent (versus human) on individuals’ propensity to tell material egoistic and social egoistic lie, respectively, and the mediating roles of moral disengagement and perceived emotional ability. Study 3 (3a and 3b) further enhanced the robustness and generalizability of the key findings. Findings Consumers are more likely to tell material egoistic lies in the face of the AI agent (versus human) because of increased moral disengagement. By contrast, they are less likely to tell social egoistic lies to the AI agent (versus human) due to the reduced perception of emotional ability of the AI agent. Potential alternative explanations are tested and ruled out. Research limitations/implications Consumers’ lying behavior toward the AI agent − an increasingly ubiquitous phenomenon in the marketplace − is not yet well understood. This research extends prior literature by proposing an integrated framework for understanding the impact of the AI agent (versus human) on consumer lying behavior and by highlighting the crucial moderating role of the type of lie (material egoistic lie versus social egoistic lie). Practical implications This study provides practical insights into effective implementation of AI agents to mitigate consumer lying. The results suggest that consumers are more likely to tell material egoistic lies to AI than to human. To mitigate such risks, firms should ensure that human employees are present in material-reward-related contexts, such as loan services, insurance claims and warranty claims. The results also reveal that, in contrast, consumers are less likely to tell social egoistic lies to AI than to human. Therefore, the deployment of AI agents should be encouraged in social-reward-related contexts, such as health-care diagnosis, performance assessment and the collection of private consumer data. The findings also inform the design of AI with respect to its intended purposes. Originality/value This research sheds light on the impact of the AI agent on consumer dishonesty. Specifically, this research contributes to the emerging literature on consumer lying, human–AI interaction and consumer self-control, and provides practical insights for more effective deployment of AI agents in the marketplace.